INDEX

Explanations

phrases related to unpopular topics or concepts

oai_token-act-pair · gpt-3.5-turbo

references to unpopularity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ural

-0.76

hner

-0.73

chn

-0.73

tein

-0.72

Vision

-0.68

sic

-0.67

ilt

-0.66

phasis

-0.66

dress

-0.66

AUT

-0.65

POSITIVE LOGITS

 unpopular

1.48

ity

0.97

 incumbent

0.89

 disadvant

0.86

 taboo

0.86

 burdens

0.85

 advoc

0.82

popular

0.80

odox

0.79

 Mehran

0.79

Activations Density 0.005%

phrases related to unpopular topics or concepts

references to unpopularity

No Comments

No Known Activations